Fourmilog: None Dare Call It Reason

« January 1, 2008 | Main | January 3, 2008 »

Wednesday, January 2, 2008

Reading List Content Management System Updated

The home-brew Perl content management system I use to maintain the Reading List Web tree has been revised and all documents it generates updated accordingly. If you're using this program to manage your own reading list, you may wish to download and install the new release. Principal changes are:

• Author and topic names which contain ISO 8859-1 accented characters are now sorted as if the characters were not accented.
• External documents may link to the frame-based reading list document and specify which page should be displayed in the main window by appending a “query” to the URL. Typical queries are:

  ?month=2006-08
  ?year=2004
  ?book=581
  ?author=Verne_Jules
  ?topic=science_fiction
  ?title=V
  ?all
  ?recent
  

  These queries automatically display the corresponding navigation panel in the sidebar and, where appropriate, scroll it with the selected item at the top. The “title” query scrolls to books with titles that begin with the first letter specified. The queries require the user's browser to have JavaScript enabled; otherwise they are silently ignored and the default entry page is displayed.
• Links generated by the <Amazon> macro have been updated to conform to Amazon's latest template for associate links.
• The <Amspec> macro has been superseded by a new <Amspecial> macro which allows general references to lists, series, search results, and other queries on Amazon, with the associate ID appended automatically in the correct form for the generated URL.
• A new <Book> macro generates links to books in the list from their permalink numbers. These links are transformed into external links when transcribed to the weblog.txt file.
• XHTML generated for the weblog.txt document automatically transforms <Month> macros from internal references within the reading list tree to external queries which display the month specified by the argument. Text from this document may now be pasted directly into a Web log posting without modification.
• Several formatting errors were corrected in the generation of the noframes.html document, and <Month> macros within it are transformed to intra-document fragment references.
• Averted the “doomsday 2009” crisis where our clever algorithm for assigning background colours to year titles would have resulted in white-on-white text.
• Metadata have been added to improve the indexing of generated documents by search engine “Web crawlers”. Each individual book item now contains a keywords declaration containing its title and author(s) (unique words only, with common articles and prepositions elided), and a robots declaration of index. The master title index serves as the access point to this document, with a robots declaration of noindex,follow. All other generated documents instruct robots to noindex,nofollow, which should prevent redundant indexing of content included in topic, author, month, and year pages.
• Several hard-coded references to Fourmilab-specific Web addresses have been replaced with configurable parameters at the top of the program.

In addition to these changes in the Web tree generation process, I have also revised all of the book entries back to the start in January 2001 to use XHTML text entities for special characters such as quotes, dashes, and ellipses, and fixed a number of typographical errors and instances of infelicitous formatting I came across in the editing process.

Starting with January 1st, 2008, and the item below posted on that date, I will cite thirteen digit ISBN-13 references (for example 978-0-375-75875-1) for all books in the reading list, even if the book was published prior to the mandate of ISBN-13 in 2007. Any ten-digit ISBN-10 can be expressed as an ISBN-13 by prefixing it with the EAN “Bookland” prefix of 978, then recomputing the final check digit with the EAN/ISBN-13 algorithm. An ISBN-13 which begins with 978 can be transformed back to ISBN-10 by the inverse process. Note that ISBN-13s which begin with the new Bookland code of 979 cannot be expressed as ISBN-10s. Books published prior to the adoption of ISBN will continue to be cited by their Library of Congress Catalogue Number, British Library Number, or other catalogue identification.

For some incomprehensible reason, Amazon.com does not accept ISBN-13 specifications in associate links. When presented with an ISBN-13, you must convert it to an “ASIN” (Amazon Standard Identification Number) which, for a 978-prefix ISBN-13, is identical to the ISBN-10 it encodes. But since 979-prefix ISBN-13s cannot be programmatically converted, there's no alternative to looking up the book on Amazon, noting the ASIN, and manually specifying it in your link. Fortunately, almost all existing ISBN-13s are of the 978 flavour (I've yet to encounter a 979), so for the moment they can be automatically interconverted as required. I've written a little Perl program to convert back and forth between ISBN-10 and ISBN-13 values, which you're welcome to download; it preserves embedded delimiters within ISBN values and also displays them in canonical form with all delimiters removed for applications (hello, Amazon!) which require them thus.

Posted at 01:18